Evaluation of satellite‐estimated surface longwave radiation using ground‐based observations

This study evaluates four satellite‐estimated data sets of surface longwave radiation (LW), the Global Energy and Water Cycle Experiment‐Surface Radiation Budget (GEWEX‐SRB), the International Satellite Cloud Climatology Project‐Flux Data (ISCCP‐FD), the Clouds and Earth Radiant Energy System‐Gridded Radiative Fluxes and Clouds (CERES‐FSW), and a newly developed data set from the Moderate Resolution Imaging Spectroradiometer (MODIS), with ground measurements at 15 sites of 2003. GEWEX‐SRB (1°*1°) and ISCCP‐FD (280 km*280 km) irradiance are available at 3‐hourly intervals, CERES‐FSW (1°*1°) and MODIS (1 km*1 km) irradiance are available at satellite overpass time. Evaluation is carried out separately for clear‐sky, cloudy sky and all‐sky conditions. Results show that, under all‐sky conditions, downwelling LW (LWDN) in ISCCP‐FD is overestimated while in CERES‐FSW it is underestimated; but for LWDN in GEWEX‐SRB and upwelling LW (LWUP) in all data sets, estimation errors are region‐dependent: no uniform trend can be found. Under clear‐sky conditions, LWDN in most data sets is underestimated, with some exceptions; LWUP in GEWEX‐SRB and ISCCP‐FD is overestimated while in CERES‐FSW and MODIS it is underestimated with some exceptions. Under cloudy sky conditions, LWDN is overestimated in GEWEX‐SRB and ISCCP‐FD while it is underestimated in CERES‐FSW; LWUP in most data sets is underestimated, with some exceptions. All data sets perform better when estimating LWUP than LWDN, during the nighttime than during the daytime, and under clear‐sky conditions than under cloudy sky conditions. However, there are substantial errors in some regions, such as LWDN overestimation by ISCCP‐FD on the Qinghai‐Tibetan Plateau. Such errors are related to the errors of input precipitable water vapor, temperature, and elevation differences.